Research On GA Overlapped Peak Analysis Algorithm Based On Previous Experience

In recent years,near infrared spectroscopy has been widely used in food,medicine,agricultural and animal husbandry products,chemical industry and other fields,making it an indispensable analytical method.With the continuous research in various fields,the complexity of the signal to be analyzed becomes higher and higher.Due to factors such as experimental conditions or substances themselves,the spectral peaks will overlap and the overlapping signals contain complex samples with various components If the multi-component wavelengths are similar,the peak is also prone to overlap,which has a greater impact on subsequent analysis.If we only rely on increasing the resolution of the instrument and improving the separation conditions,the overlapping peaks can hardly be separated completely.As the complexity of the components increases,the analysis and precise quantification of overlapping peaks become the important contents of current research.In this paper,the traditional overlapping peak analysis algorithm is introduced first,and its advantages and disadvantages are analyzed.Typical methods include Fourier transform,wavelet transform,artificial neural network and curve fitting.The results show that the above method has some effects on the analysis of overlapping peaks.However,the complexity of the components to be measured is getting higher and higher,and the degree of overlap increases,making the traditional methods have some problems in the analysis,for example,large amount of calculation,There are more people to imagine,errors and so on.Based on the analysis of the existing overlapping peak analysis methods,this paper presents a new method to resolve overlapping peaks,that is,the GA algorithm which combines the previous experience.Genetic algorithm is a parallel stochastic search optimization method that simulates the evolution of nature.It has high robustness and adaptability.In this paper,the wavelet threshold is selected to deal with the noise in the original signal,the white noise and the sinusoidal noise in the signal can be well processed to get the smooth spectrum.The initial parameters of the overlapped peak are predicted by the second order differential.According to the genetic algorithm Global search ability,the use of genetic algorithms to optimize the Gaussian model and peak-related parameters to obtain the optimal peak results.A large number of experiments show that the overlapped peak analysis method proposed in this paper can effectively improve the resolution of overlapping peaks.By using the Gaussian function fitting overlap and multi-peak analysis,the comparison of the simulated spectrum and the original spectrum shows that the method has good reliability and will be used as a promising new method.